Skip to content
GET/api/v1/monitoring/drift

Get model drift detection

Compares the current operational metrics snapshot against the org's stored drift baseline and returns per-metric drift results plus a severity summary { total, critical, high, medium, low, none, hasDrift }. The ten metrics compared are latency p50/p95/p99/mean, error_rate, token_usage total/prompt/completion, cost_usd and request_count — operational drift only; there is no input-distribution, output-distribution, per-model or per-prompt analysis. Detection is a z-score test (threshold 2.0, min 3 samples). The org is resolved from ?orgId= → body.orgId → the authenticating eg_ key's own org, and membership is verified when one is named; a caller with no resolvable org receives an empty result rather than another tenant's. Caveat: the metrics snapshot is a process-wide aggregate and the baseline lives in process memory (max 500 orgs, oldest evicted) and is lost on redeploy.

Authentication

Send Authorization: Bearer YOUR_API_KEY on every request. Generate API keys at /dashboard/settings/api-keys.

Response

200 example

{
  "success": true
}

All status codes

200Drift signals.
400(no description)
401(no description)
403Forbidden — the credential is valid but lacks the API-key scope, member role, or plan entitlement this operation requires.
429(no description)
500Internal Server Error — an unhandled error was converted to the standard error envelope (INTERNAL_ERROR).

Code samples

cURL

curl -X GET \
  https://evalguard.ai/api/v1/monitoring/drift \
  -H "Authorization: Bearer $EVALGUARD_API_KEY"

TypeScript

// The TypeScript SDK (@evalguard/sdk) exposes TYPED methods — runEval,
// getEval, runSecurityScan, checkFirewall, … — not a generic request().
// For an arbitrary endpoint, call it directly:

const res = await fetch("https://evalguard.ai/api/v1/monitoring/drift", {
  method: "GET",
  headers: { Authorization: `Bearer ${process.env.EVALGUARD_API_KEY}` },
});
console.log(res.status, await res.json());

Python

# The Python SDK (pip install evalguardai) exposes TYPED methods on
# EvalGuardClient — run_eval, get_eval, … — not a generic request().
# For an arbitrary endpoint, call it directly:

import os
import requests

headers = {"Authorization": f"Bearer {os.environ['EVALGUARD_API_KEY']}"}

response = requests.request("GET", "https://evalguard.ai/api/v1/monitoring/drift", headers=headers)
print(response.status_code, response.json())

Go

package main

import (
	"context"
	"fmt"
	"net/http"
	"os"
)

func main() {
	req, _ := http.NewRequestWithContext(context.Background(), "GET", "https://evalguard.ai/api/v1/monitoring/drift", nil)
	req.Header.Set("Authorization", "Bearer "+os.Getenv("EVALGUARD_API_KEY"))
	resp, err := http.DefaultClient.Do(req)
	if err != nil { panic(err) }
	defer resp.Body.Close()
	fmt.Println(resp.Status)
}

Errors

400401403429500

Other Monitoring endpoints